Brain Tumor Detection and Visualization in VRAR

Abstract

Globally, cancer is the second leading cause of death. Approximately 70% of deaths from cancer occur in the lower middle-income countries. Head and neck cancers are the sixth most common cancers worldwide, with 630000 new cases diagnosed annually, causing 350000 deaths. [1] Globally, brain tumors are also a significant source of cancer-related morbidity and mortality, with an overall incidence of 4–5/100 000 cases annually, contributing to 2% of all cancer deaths, ranking at 10th place among cancers as the leading cause of death. [2] [3] In Pakistan, brain cancers rank at 11th place, where 150000 new cases of cancer are diagnosed annually, causing 60%–80% of deaths. Most deaths and grievous side effects arise from the lack of tumor detection in time when the optimal time for treatment has already passed. This may be because of a multitude of reasons, like deficiency of money, doctors, and the belief that such a thing cannot occur to oneself. We proposed a solution in the form of Computer-Aided Diagnostic (CAD), where MRI images would be classified and tumors segmented by convolutional neural network (CNN) models, trained on previously given data sets. This would serve as an assisting tool for doctors to get a second opinion of sorts and to minimize any human error that may result in a false-negative result, particularly in a situation where false-positives are highly preferable to false negatives. Our solution would also allow any person who had an MRI of their brain done, for any other reason can process their MRIs to ensure that no tumor has been formed. This easy and cheap solution to detect tumors would decrease the average time between the formation of a tumor and its detection. We further aim to make a 3D model, VR/AR compatible, that would allow you to easily visualize and manipulate the tumor inside the brain model. This would allow for an understanding that a plain 2D image may not convey. The model could be used as training material to train new doctors, as well as allow patients to be better informed due to easy-to-understand visuals. The model can also be used to trace the progression of a patient’s tumor, allowing the doctors and medical staff to enhance their understanding of the tumor, and thus propose the optimal treatment plan.